Circulating Spike Protein Detected in Post–COVID-19 mRNA Vaccine Myocarditis

Sixty-one adolescents and young adults between 12 and 21 years of age, including 16 individuals with vaccine-associated myocarditis, provided a blood sample for analysis (Figure 1). Demographics are described in the Table 1. The majority of individuals with postvaccine myocarditis were male (n=13 of 16; 81%), and symptom onset typically occurred within the first week after vaccination (median, 4 days; range, 1–19 days). In the postvaccine myocarditis cohort, most of the individuals (n=12 of 16; 75%) developed myocarditis after the second dose, although 2 experienced symptoms of myocarditis after the first dose and 2 after the third booster dose. All patients presented with chest pain, and all were found to have elevated cardiac troponin T (median, 260 ng/L; interquartile range, 215–1114 ng/L; upper limit of normal, 14 ng/L) and C-reactive protein (CRP) levels (median, 29.75 mg/L; interquartile range, 15.53–50.58 mg/L; upper limit of normal, 8 mg/L; Table S1). Samples were also collected from 45 age-matched asymptomatic vaccinated control subjects up to 3 weeks after the second mRNA vaccination.

Given the desired goal of humoral protection against SARS-CoV-2 infection, serological responses were compared between the individuals who developed postvaccine myocarditis and age-matched, asymptomatic vaccinated control subjects. Because antibody responses are dynamic over time, only samples collected within 11 days after the second vaccine dose were analyzed to allow a more evenly matched comparison of antibody responses (myocarditis cohort, n=10; healthy vaccinated control subjects, n=17). In this subgroup, no differences were seen in anti-spike or anti-RBD (anti-receptor binding protein) immunoglobulin (Ig) M, IgG, or IgA levels (Figure 2A and 2B). To more deeply assess potential qualitative differences in the serological responses associated with postvaccine myocarditis, we evaluated the ability of the antibody to engage Fc receptors and additional Fc effector functions. No overall difference in binding to FcγR2a, FcγR2b, FcγR3a, or FcγR3b was observed (Figure 2C). Consequently, opsonophagocytic and complement-activating potentials were comparable between the groups (Figure 2D). Collectively, these data indicate that individuals who developed myocarditis have a humoral immune response comparable to that of asymptomatic adolescents and young adults (Figure 2E). We found no indication that a specific antibody response is associated with myocarditis, but instead, all adolescents and young adults mounted a substantial immune response, conferring protection against SARS-CoV-2 after vaccination.

In addition to vaccine-related responses, we looked more broadly at the antibody response in a subset of patients and vaccinated control subjects by performing phage immunoprecipitation sequencing for self-antibodies (Figure 2F) and a phage immunoprecipitation sequencing–based VirScan, a phage display approach for in-depth characterization of antibodies to the human relevant virome (Figure 2G). No significant levels of self-antibodies were noted in the myocarditis group compared with the healthy vaccinated control group. Strong antibody responses to common pathogens, including other respiratory viruses (respiratory syncytial virus, influenza, and β-coronaviruses), herpes viruses (herpes simplex virus type 1, cytomegalovirus, and Epstein-Barr virus), and vaccine strains (measles, rubella, and mumps), were detected. However, no specific difference tracking with the development of myocarditis stood out in this limited cohort.

We then sought to characterize T-cell responses by multiparameter flow cytometry. We first compared the distribution of the different memory subsets, including all CD4⁺ and CD8⁺ T cells, between a subset of individuals who developed postvaccine myocarditis and asymptomatic vaccinated control subjects using age- and sex-matched samples with comparable time from vaccination. Overall, there were no differences in the frequencies of naive (CD45RA⁺CCR7⁺), central memory (CD45RA⁻CCR7⁺), effector memory (CD45RA⁻CCR7⁻), and terminally differentiated effector memory (CD45RA⁺CCR7⁻) T cells between the 2 groups except that individuals with myocarditis had slightly higher frequencies of effector memory cells (t test, P=0.047; Figure S2A and S2B). We also did not see a difference in the proportions of memory subsets in SARS-CoV-2 spike–specific CD4⁺ T cells (Figure S2C). Furthermore, the frequencies of interferon-γ secreting and degranulating, as determined by CD107a expression, CD8⁺ T cells, and CD4⁺ T cells on stimulation with SARS-CoV-2 spike peptides, were indistinguishable between both groups (Figure S2D–S2G). The only noticeable difference in the T-cell signature was a higher frequency of PD-1–expressing bulk CD4⁺ T cells in the individuals with postvaccine myocarditis (unpaired t test, P=0.02; Figure S2H and S2I), likely reflecting variability in expansion after immunization but potentially also suggesting a higher level of exhaustion in this cell subset.

Although minimal differences were seen in the T-cell populations, individuals with myocarditis displayed distinct cytokine profiles reminiscent of the profile seen in MIS-C,^17–19 suggesting likely innate inflammatory activation, with significantly elevated levels of interleukin (IL)-8, IL-6, tumor necrosis factor-α, IL-10, interferon-γ, and IL-1β and lower IL-4 levels compared with healthy vaccinated control subjects (Figure 3A–3G) with no notable differences in IL-5, IL-12p70, or IL-22 (Figure 3H–3K). To further characterize the inflammatory profile, complete blood counts were analyzed. Although mostly within normal ranges in both cohorts, total leukocytes, specifically neutrophils, were significantly increased in individuals with postvaccine myocarditis compared with vaccinated control subjects (unpaired t test, P=0.007 and P=0.01, respectively), whereas platelet counts were decreased compared with vaccinated control subjects (unpaired t test, P=0.03; Figure S3). These results suggest that postvaccine myocarditis is associated with normal adaptive and T-cell immunity but modest innate activation.

To further characterize immune responses and to assess for potential stimuli of innate inflammation, ultrasensitive single-molecule array antigen assays were used to measure the vaccine-stimulated production of both the full-length SARS-CoV-2 spike protein and its cleaved subunit, S1, in the blood after intramuscular vaccination. Given that S1 antigen levels were seen to drop quickly in a healthy adult cohort, we sought to assess the clearance of S1 in our adolescent cohorts. As seen in adults, in analyses of adolescent plasma samples, which were collected primarily after their second vaccination, freely circulating S1 was not detected in most of the samples (Figure 4A).

We then sought to determine whether S1 was undetectable because the antigen had been effectively cleared or because the antigen was bound and masked by circulating antibodies. To release antibody-bound antigen, we first treated plasma samples with dithiothreitol to denature the antibodies. After incubation with dithiothreitol, S1 was detected in plasma of 34% of the vaccinated control cohort and 29% of the myocarditis cohort. We repeated this analysis for the adult cohort, but S1 antigen remained undetectable in adults after the second vaccine dose (Figure S4). Although the adult cohort is small (n=13), these findings may suggest an age-related difference in the processing of the mRNA vaccine. Regardless, the frequency with which we could detect S1, as well as the similar mean concentrations between the 2 adolescent cohorts, suggests the findings are therefore unlikely associated with the development of postvaccine myocarditis.

In notable contrast, adolescents who developed myocarditis had markedly higher levels of free full-length spike protein in their plasma (33.9±22.4 pg/mL), unbound by antibodies (Figure 4A), whereas asymptomatic vaccinated control subjects had no detectable free spike protein (unpaired t test, P<0.0001). Although postvaccine myocarditis clinically occurs more commonly in males, elevated spike was seen equally in both affected females and males (Figure S5). We found a minimal increase in spike levels after dithiothreitol treatment, suggesting that most of the antigen is freely circulating and unbound by antibodies in the individuals with postvaccine myocarditis. Antibody-bound spike was detected in only ≈6% of the vaccinated control subjects. Similarly, we were not able to detect any free or antibody-bound spike in the healthy adult samples (Figure S4).

When analyzed according to time since vaccination, free S1, which was detected in only one patient with postvaccine myocarditis and one vaccinated control subject, was detected only within the first week. However, antibody-bound S1, which was detected in roughly one-third of both cohorts, could be detected up to 3 weeks after vaccination (Figure 4B). In contrast, both free and antibody-bound spike, which was detectable only in patients who developed vaccine-induced myocarditis, remained detectable up to 3 weeks after vaccination (Figure 4B). Longitudinal sampling displays a slow decline in both free and antibody-bound spike, suggesting that collecting blood at only a single time point was unlikely to miss circulating spike in healthy vaccinated control subjects. To assess whether this persistence of spike in patients with myocarditis was attributable to inadequate antibody neutralization, we carried out in vitro single-molecule array neutralization assays.²⁰ We analyzed plasma samples from a subset of patients with myocarditis (n=9) and healthy vaccinated control subjects (n=14) for whom plasma samples were collected within 1 week after the second vaccine dose; however, no significant differences in antibody neutralization capacities were observed (Figure S6).

The persistence of circulating spike in patients with postvaccine myocarditis is similar to the SARS-CoV-2 antigenemia previously reported to be a pathogenic feature of MIS-C.^18,21 For that reason, we compared S1, spike, cardiac troponin T, and CRP levels between individuals who developed myocarditis and those who developed MIS-C with cardiac complications (Figure 5). There were no significant differences in the mean S1 and spike concentrations between the 2 groups. Although both MIS-C and postvaccine myocarditis resulted in elevated cardiac troponin T and CRP compared with healthy vaccinated control subjects, when MIS-C was compared with postvaccine myocarditis, cardiac troponin T levels were significantly elevated for the myocarditis cohort, and CRP levels were significantly elevated for the MIS-C cohort.

For a subset of patients with postvaccine myocarditis (n=4) from whom we obtained repeat blood sampling, we generated longitudinal immunoprofiles, including cardiac troponin T, S1, spike, anti–SARS-CoV-2 antibody, and cytokine levels (Figure S7). Although spike levels can remain elevated for several days to weeks for all patients, any detected S1 was generally cleared. In all 4 patients, anti-N IgG was undetectable, suggesting that natural infection with SARS-CoV-2 was unlikely a contributing factor. In contrast, anti-spike, anti-S1, and anti-RBD IgG, IgA, and IgM levels increased as expected after vaccination. Patients 1 through 3 developed myocarditis after the second vaccine dose; patient 4 developed myocarditis after the first dose. Of the cytokines, IL-8 was most prominent, mirroring cardiac troponin T and antigen levels most closely.

Although epidemiological reports describe key clinical features associated with myocarditis after vaccination with BNT162b2 or mRNA-1273,^9,22 here, we provide in-depth immunoprofiling of patients with postvaccine myocarditis. We discovered that individuals who developed postvaccine myocarditis uniquely exhibit elevated levels of free spike protein in circulation, unbound by anti-spike antibodies, which appear to correlate with cardiac troponin T levels and innate immune activation with cytokine release. However, adaptive immunity and T-cell responses were essentially indistinguishable from those of asymptomatic age-matched vaccinated control subjects. The postvaccine myocarditis immunoprofile is distinct, however, from acute SARS-CoV-2 infection and the delayed postinflammatory illness MIS-C. Although these findings might provide insight into the immunophenotype of vaccine-related myocarditis, they do not alter the risk-benefit ratio strongly favoring vaccination¹⁰ to protect against severe COVID-19–related complications.

It is notable that spike, which remained intact by evading cleavage and clearance, was associated with myocarditis in this cohort. Whether the circulating spike protein in the setting of mRNA vaccination was pathogenic is unclear. In postvaccine myocarditis, the spike protein appears to evade antibody recognition because the anti-spike antibodies that are generated are produced in adequate quantities with normal functional and neutralization capacity. There is growing in vitro evidence that spike itself can stimulate cardiac pericytes dysfunction²³ or inflame the endothelium, potentially by downregulating angiotensin-converting enzyme 2 expression, by impairing endothelial nitric oxide bioavailability,²⁴ or by activating integrin-mediated inflammation with hyperpermeability of the endothelial cell layer.²⁵ Thus, the spike antigen itself, which evades antibody recognition rather than invoking immune hyperactivation, may contribute to myocarditis in these individuals.

T-cell subsets have been implicated in other forms of myocarditis, such as autoimmune myocarditis and dilated cardiomyopathy.²⁶ Although we did not observe any difference in SARS-CoV-2–specific CD4⁺ and CD8⁺ T cells, nor did in vitro stimulation reveal any differences in interferon-γ response between the 2 cohorts, we cannot rule out T-cell involvement on the basis of this analysis alone. However, a slight increase in PD-1–expressing bulk CD4⁺ T cells in postvaccine myocarditis may indicate some degree of T-cell activation or exhaustion, although it appears distinct from SARS-CoV-2–specific responses. In addition, although we did not detect any autoantibodies as has been seen in other forms of myocarditis, this does not necessarily rule out the potential for T-cell involvement. However, our data suggest that there is a potentially unique mechanism of myocardial injury after SARS-CoV-2 mRNA vaccination associated with a robust innate immune response. Future studies are needed to fully investigate the contribution of T cells in postvaccine myocarditis.

Vaccine-related myocarditis is not unique to the mRNA vaccines; myocarditis can also occur after other vaccines,^27,28 including non–mRNA COVID-19 vaccines, such as influenza and smallpox vaccines. Therefore, it is possible that circulating spike is a biomarker of immune dysregulation leading to myocarditis rather than a causal agent. However, we discovered distinct differences in how adolescents respond to mRNA vaccination compared with adults, which warrant further investigation. It has previously been shown that after the first inoculation of the mRNA-1273 vaccine, the cleaved S1 subunit of spike can be detected in the plasma of healthy adults.¹⁶ However, after the second dose, no antigen was detected,¹⁶ presumably because there are higher levels of circulating anti–SARS-CoV-2 antibodies, which quickly bind any circulating antigen, facilitating its clearance. In contrast, one-third of the adolescents displayed antibody-bound S1 antigenemia after the second vaccination, regardless of the development of myocarditis, a finding not seen in our smaller sample of adults. This suggests that either the immune system of adults responds more quickly to the vaccine-induced production of spike or, because of differences in body mass, the levels of S1 fall below the limit of detection for adults. Alternatively, increased levels of free spike compared with free S1 may be attributable to differences in renal clearance rates; S1 would be expected to clear faster with a molecular weight of 76 kDa, approximately half that of full spike (180 kDa). Because both adults and the adolescents included in our cohort received adult dosing of the mRNA vaccine, this finding suggests an age-related capacity for handling vaccine-introduced antigen. It is important to note that the majority of circulating S1 was bound by specific anti-S1 antibodies, indicating an appropriate immune response for targeting and clearing S1.

Identification of SARS-CoV-2 viral particles in the blood is not unique to postvaccine myocarditis. However, when antigenemia is detected in severe COVID-19²⁹ or MIS-C,^18,21 a more pronounced hyperinflammatory, superantigen-like response is typically seen, with spike immune complexes triggering hyperactivation of monocyte phagocytosis³⁰ and neutrophil extracellular trap formation.³¹ However, in vaccine-induced myocarditis, spike does not appear to be bound to antibodies, and there is no evidence of excessive immune complex activation of neutrophils, complement, or monocytes. In addition, CRP levels are markedly elevated in MIS-C compared with postvaccine myocarditis.³² This distinction in hyperinflammation may result from the source of antigenemia. In vaccine-associated myocarditis, antigenemia results after sterile intramuscular inoculation with the liposomally protected mRNA transcript. In contrast, in MIS-C, antigens may leak into the circulation as a result of dysbiosis in the gut, which signals the zonulin-mediated loss of tight junctions, as evidenced by elevated zonulin levels, lipopolysaccharide-binding protein, and soluble CD14.¹⁸ It is also possible that associated endotoxemia exacerbates the inflammatory response against SARS-CoV-2 antigens, accounting for the severe hyperinflammatory response characterizing MIS-C. Furthermore, in MIS-C, the superantigen-like motif³³ of spike/S1 engages with the immune system to produce T-cell receptor skewing³⁴ and a profound hyperinflammatory response³⁵; in postvaccine myocarditis, the circulating free spike antigen appears to evade antibody recognition. Correspondingly, post-mRNA vaccine myocarditis follows a more subdued acute course,^8,36 but long-term outcomes remain to be seen.

Limitations of this study include the relatively small sample size because postvaccine myocarditis is a rare complication, with ≈18 cases occurring for every 1 million vaccine doses administered. Our cohorts were not evenly balanced between the BNT162b2 and mRNA-1273 vaccines: all of our adolescent control cohort and the majority of our myocarditis cohort received the BNT162b2 vaccine (n=15). Furthermore, a more extensive analysis of T-cell subsets, including Th17 cells, which have previously been implicated in non–vaccine-induced myocarditis/dilated cardiomyopathy,²⁶ or cardiac biopsy tissue analysis would be informative in the future. Nonetheless, our multicenter collaboration allowed meaningful analysis of a considerably sized postvaccine myocarditis cohort. In addition, ultrasensitive antigen detection through single-molecule array²⁹ does not distinguish whether spike antigenemia is the cause or consequence, and not all patients with myocarditis had detectable antigenemia. Rather, immunophenotyping lays important groundwork to further understand mRNA vaccine–associated complications and provides rationale for future research to aid in vaccine design and dose. These findings also suggest that administration of anti-spike antibodies, if spike antigenemia is detected, could potentially prevent or reverse postvaccine myocarditis.

It is reassuring that neither the BNT162b2 vaccine nor the mRNA-1273 vaccine induces abnormal adaptive immunity or T-cell responses associated with immune activation targeting the myocardium. However, we detected free spike antigen in the blood of adolescents and young adults who developed post-mRNA vaccine myocarditis. Although the implications of this finding must be better understood, these results do not alter the risk-benefit ratio favoring vaccination against COVID-19 to prevent severe clinical outcomes.

The authors thank the children and families who participated in this research. They thank Jeni Melo, BS, of Boston Children’s Hospital for her assistance in collecting clinical data. Conceptualization and methodology: L.M.Y., Z.S., Y.C.B., G.A., A.F., D.R.W. Investigation: L.M.Y., Z.S., Y.C.B., M.B., A.K., J.P.D., M.L., B.P.B., T.N., Y.S., C.-A.C., B.J., A.G.R., G.A., A.F., D.R.W. Formal analysis and interpretation: L.M.Y., Z.S., Y.C.B., C.-A.C., A.G.E., J.C., A.D., D.B., M.L.-R., M.A., B.J., A.G.R., G.A., A.F., D.R.W. Writing: L.M.Y., Z.S., Y.C.B., G.A., A.F., D.R.W. Writing–review and editing and supervision: L.M.Y., Z.S., Y.C.B., C.-A.C., A.G.E., J.C., A.D., D.B., M.L.-R., M.A., B.J., A.G.R., G.A., A.F., D.R.W.